12-09-2024
How do you optimize the harnessing of solar energy, build better distillation trains, or design more efficient network systems?
Those are some of the challenges being studied by Mohit Tawarmalani, executive associate dean of faculty and Allison and Nancy Schleicher Professor of Management at the Mitch Daniels School of Business. His research at the intersection of computer science, optimization, and operations management has won several awards, most recently the Computing in Chemical Engineering Award from the Computing & Systems Technology Division of the American Institute of Chemical Engineers, an honor he shared with Rakesh Agrawal, Winthrop E. Stone Distinguished Professor of Chemical Engineering.
As a result of his expertise in the field, Tawarmalani was invited by Purdue President Mung Chiang to present at the Westwood Lecture Series, an exclusive gathering of faculty engaged in a particular research topic. The goal of the series is to enhance the intellectual vibrancy of the campus.
Tawarmalani’s talk was “Optimization Beyond Convexity: Applications in Process Design, Networking and Pricing.” He spoke about the fault line between convexity and non-convexity, a delineation that divides tractable problems from treacherous ones. He explored optimization algorithms for nonconvex models from disparate sources including consumer behavior, unpredictability of network state, and physical phase behavior, which play an important role in product pricing, resiliency of operations, and design of sustainable processes. Tawarmalani also discussed unifying advances in treating nonlinear models involving products and ratios of decision variables that lead to insights and solutions for vertical differentiation and operations.
“We had a robust conversation about a number of topics, such as process design, resiliency, and pricing,” Tawarmalani says. “By using a divide-and-conquer approach, where you use rigorous approximations to make the search for solutions to very complex problems easier by breaking them into smaller sub-problems, they become more manageable.
“The solutions have real impact. New distillation configurations can provide almost 15% in energy savings, equivalent to 245,000 barrels of oil a day, and 16% reduction in C02 equivalent. Companies like Google and Meta need these optimization techniques to provide proper certifications and service levels for their wide area network operations.”
Tawarmalani is academic director of Purdue’s Krenicki Center for Business Analytics and Machine Learning. He was a founding co-director of the university’s Master of Science in Business Analytics and Information Management program and co-chair of the design committee for the undergraduate degree in Business Analytics and Information Management. He also led the Purdue team that won the INFORMS UPS George D. Smith Prize for innovative educational practices in training students to be practitioners of operations research and analytics.
